Page 8 - i1052-5173-27-9

mapping interface. We began the study
with conventional 2D mapping, albeit
aided by 2D digital mapping techniques
using the GIS data structure of Pavlis et al.
(2010) with QGIS software. Orthocorrected
satellite imagery with resolutions of 1–2 m
from the USGS and ArcGIS online were
used as a base map for our 2D mapping.
ArcGIS Pro and Midland Valley’s Move
software were used for draping 2D map
data onto the USGS DEM for the area and
for comparison with 3D mapping results
(see Brush, 2015, for more details on
workflows).
Results
Our 2D geologic map is high resolution
by almost any standards due to GPS posi-
tioning and the resolution of the orthoim-
agery (Data Repository supplement 2 [see
footnote 1]). Nonetheless, the problems of
conventional 2D mapping and 2.5D drap-
ing to a terrain model quickly became
apparent when we attempted to analyze the
data in 3D.
Figure 2 shows a comparison of a 2.5D
image and linework drapes (2A) onto a
low-resolution elevation model versus
mapping directly onto a true 3D view
afforded by the SfM models (2D). The
principal source of the distortions in
Figure 2 include (1) artificial smoothing
of the terrain in the low-resolution model
leading to errors in elevation positions
of image pixels, which transfer to the
geologic interpretation; and (2) errors
inherited from the orthophoto production
process that are transferred to the image
drape.
Beyond these issues of spatial errors
from the 2.5D method, we suggest that the
greatest strength of SfM 3D surface mod-
els is the increased geologic insight that
can be gained from using these distortion-
free, 3D visualizations as a mapping base
during and after fieldwork. Probably every
field geologist has wanted the ability to
“fly like a bird” to view features from dif-
ferent perspectives. Indeed, this is one
reason helicopters are used in field studies
and is the most obvious reason UAS are
beginning to see widespread use in field-
work (e.g., Jordan, 2015; Hackney and
Clayton, 2015). SfM models UAS west-vergent recumbent fold that was
flight video provide a virtual experience
close to this capability at a tiny fraction of the second cleavage (Pavlis et al., 2016).
the cost of a helicopter and allow limitless
virtual views of the scene that is impossible
8 GSA Today | September 2017
from a live aircraft flight. As an illustra-
tion of the power of this capability, Figure 4
shows the evolution of our understanding
of the structure shown in Figures 2D and
2E. In the 2D field map from the initial
visit to the site (Fig. 4A), the field interpre- steep terrain, or both. Thus, how many
tation was relatively simplistic and contra-
dictory. Field-note sketches considered
several alternative fold geometries, but the
initial work was inconclusive. In a second
visit to the site, more orientation data were
obtained and photographs were acquired
for the SfM model shown in Figure 2. Like
the first visit, however, multiple hypoth-
eses were considered for this structure, and
a field sketch (Fig. 4C) at the end of the
field day was the working hypothesis.
After later analysis manipulating 3D visu-
alizations and mapping onto the SfM
model, we realized that the structure was a
large, refolded recumbent fold (Fig. 2F).
This hypothesis was confirmed by a third
field visit to the site.
Clearly this was not a controlled experi-
ment and arguably we would have recog-
nized the structure anyway, either with
more field time or through traditional
methods like serial section construction.
Moreover, the approach was inefficient
relative to our present workflow model
because we were developing techniques at
the time. Nonetheless, the ease of the anal-
ysis from the 3D visualization made recog-
nition of the feature easier and led to
greater confidence in the interpretation.
Similarly, 3D analysis of the broader
area in this study answered several ques-
tions (e.g., Data Repository supplement 2
[see footnote 1]) but, perhaps more impor-
tantly, led to hypotheses that probably
would not have arisen without the 3D map-
ping. For example, directly along struc-
tural trend from Figure 2, outcrop-scale,
plunging, type 3 (coaxial) refolded folds
(terminology of Ramsay, 1967) like those
in Figure 2 are common. However, the ori-
entation of the most prominent isoclinal
folds is grossly different along strike—
approximately recumbent to the north and
upright to the south. In the absence of a 3D
model, this observation is difficult to eval-
uate, but using the 3D model to visualize
geometry across the area, our working
hypothesis is that there is a large-scale
who have only used 2D methods (maps
refolded by upright folds associated with
More work is needed to test that hypoth-
esis and will be the subject of future
fieldwork. Nonetheless, the important
point is that in the absence of the 3D visu-
alization we probably couldn’t have even
raised this question without much more
fieldwork, the ability to climb across the
other unresolved geologic problems or
missed issues lie hidden in steep terrain
that could be resolved with these methods?
3D Mapping and its Importance to 3D
Modeling
This case study gives a partial illustra-
tion of the potential of using SfM for solv-
ing geologic problems, but it is a limited
example in the broad range of potential
applications. The key features in this case
were (1) the dramatic increase in accuracy
of the 3D view, which aided confidence in
geometric interpretations as real, not arti-
facts of mapping imprecision; and (2) the
ability to view 3D features from a variety
of viewpoints, and revisit these views
repeatedly, allowing fast evaluation of
geometry, something impossible with con-
ventional mapping. This ability is a cogni-
tive breakthrough for field geology
because it allows geologists to break from
the traditional paradigm (e.g., Compton,
1985) that key features should always be
recognized the first time around due to the
economics and logistics of fieldwork—i.e.,
this paradigm may still hold for the field
visit, but a key site can now be captured as
a 3D visualization that can be viewed ad
infinitum to help resolve problems.
For those inexperienced with field geol-
ogy in areas of complex structure, particu-
larly in steep terrain, it may not be obvious
how important these abilities can be. From
our experience, the 2.5D method can be
used for construction of 3D geologic mod-
els of complex structure (e.g., Pavlis et al.,
2012), but the distortions and imprecision
in the underlying terrain model make geo-
logic model construction inefficient as well
as potentially wrong due to uncertainties
in the sources of spatial error. In a true 3D
model based on SfM, none of those spatial
uncertainties exist in the raw data, and the
only uncertainties arise from potential
interpretation errors—a problem much
more easily evaluated through an iterative
mapping approach. Note also that for those
and cross sections) for geologic analysis, it
is easy to underestimate the difficulty of
constructing a true 3D geologic model